Team Root
Language in the Future of Computing
Web Sites Reviewed
Thesis Statement
Thesis
The dominant methods of human computer interaction place several constraints in their communication producing undesired effects such as errors, and incorrect output. Natural language processing technology, as it is bettered, can rectify these problems and make computers more attractive to technophobes.
Older
"Natural language processing will shape the way humans and computers interact in the future."
Keywords
natural language processing, sentiment analysis, talking to robots, voice recognition, grammar, linguistics, sentence parsing, vocabulary, homonyms, multiple languages, machine learning, human-computing interaction, chat-bots, Google search, IBM Watson, Apple Siri, Semantics, text mining
Bibliography
Books
Grand, Steven: The Creation
Pinker, Steven: The Language Instinct
Pinker, Steven: Words, Rules and their Meanings
Web Sites
http://irevolution.net/2011/10/10/recordedfuture/
http://www.bbc.co.uk/news/technology-14841018
http://www.youtube.com/watch?v=Jd3-eiid-Uw
http://www.cs.cofc.edu/~manaris/publications/advances-in-computers-vol-47.pdf
http://www.parc.com/content/attachments/PARC_NaturalLanguage_backgrounder.pdf
Research Notes
Research Structure
Section 1: HIDs. Directionality of communication.======
Common HIDs: (Source: http://en.wikipedia.org/wiki/Human_interface_device)
- Keyboards
- Mice, Touchpads
- Cameras
- Fingerprint Scanners
- Headsets
Less Common HIDs
- Driving Simulator Devices
- Wired gloves
- Dance pads
- Wii Remote
- Surface computing devices
- Section 2: Keyboards
- Types of Keyboards
- Issues with Keyboards
- Keystroke Logging (wireless)
- Physical Injury
- Repetitive Strain Injury
- Carpal Tunnel Syndrome
- vHealth Risks
- Language
- Standards
- Size Constraints/Limited functions on mobile devices
- Common Damage
- Hick's Law
How can voice replacement software solve these problems
Section 3: Mice
- Issues with Mice
Section 4: Competition for NLP
- Touch Interfaces
- Gesture/Motion Sensors
Section 5: Supplements for NLP
- Augmented Reality
Research Notes
Introduction
Computers have become a mainstay of Human lives, being involved in several of their activities, from their jobs, to communication, to entertainment. However a usage difficulty is still attached to them and technology in general, giving birth to self-categorizations such as technophobes and technophiles. This perceived difficultly lies not at the conceptual level, but at the specific human computer interaction level. The dominant devices that are used to accomplish these tasks today include Keyboards and Mice, among others. While these devices are cheap, and in wide use, they pose several problems, which can be eliminated with better technology.
The dominant methods of human computer interaction place several constraints in their communication producing undesired effects such as errors, and incorrect output. Natural language processing technology (hence abbreviated NLP), as it is bettered, can rectify these problems and make computers more attractive to technophobes.
Human Interface Devices Today
Human computer interaction is a complex process and requires different kinds of devices for different types of interactions. There are several specialized devices built just to accomplish this same task. However this method causes several problems, such as fragmentation, hardware costs, device dependence, training, and hence errors.
The following table shows the most common HIDs in use today, and their associated average cost.
HIDs (Widely Used)
Cost
HIDs (Specialized)
Cost
Keyboards
$20
Specialized Single Purpose Devices
$150
Pointing Devices (Mice/Touch based)
$20
Wired Gloves
$300
Joysticks
$50
Dance Pads
$300
Cameras
$50
Wii Remotes
$80
Fingerprint Scanners
$70
Microphones
$20
The most common place of these of course are Keyboards and Mice, which are now a standard issue with every personal computer. Each of these devices serves a separate function and neither can completely replace the other, setting back the average PC owner by $100 per purchase. Mice in general are portable, thanks to their size, and wireless mice are very popular today with Laptop like devices.
These two devices will be the key focus of comparison to NLP devices in this study, and are explored in detail in the sections below.
Keyboards
Introduction/History
The earliest computers used typewriters (such as Teletype) as keyboards. They followed the QWERTY standard with some added keys such as 'Escape' and 'Control'. But as computers became more complex and needed more function keys, various manufacturers experimented with various new key layouts and models. Among those that did catch on at least for a while were the famous Dvorak, the Chorded, Thumbpads, Finger-Switch keyboards. None of these really made it to the mainstream, QWERY remained the ever popular. Manufacturers often make small changes to the layout to accommodate special keys specific to their machines, or otherwise sometimes for the form factor. Even today however, there are those that are passionate about Dvorak keyboards, and will use it and only it.
Types (?)
The fact that keyboards had to go through so many iterations only to end up with unsatisfactory results for so many suggests that perhaps these devices are not suited to all needs and people. In fact they carry a whole host of issues with them:
Security Issues
Keyboards are the primary form of textual input for any user. Thus any malicious software, such as keystroke loggers can record a user's strokes.
Health Risks
Keyboards are also said to expose their users to various health risks. They require a person to sit in a relatively constrained position for the entire duration of their usage. While this is alright in short doses, it can be fairly harmful in the long run. Additionally, people with back problems might not be able to use standard keyboards at all.
The required position of the hands can cause other problems as well, such as a Repetitive Strain Injury. Though the prevalence of this is known to be low, injuries such as Carpal Tunnel Syndrome are associated with heavy Keyboard use. One study that monitored a thousand individuals for development of strain in the median nerve of the wrist found 4.8% of the participants reported a strain over a year. While this number is not alarming, one must consider the short time period used in the study.
http://jama.ama-assn.org/content/289/22/2963.short
Other risks have also been described with the use of public keyboards. Another study found that the average public Keyboard carries enough germs to be able to infect a person. Offices around the world do use disinfectants and cleaners on keyboards often, however in other settings such as University/College campuses, libraries, etc, these standards are not met.
http://abcnews.go.com/Health/Germs/story?id=4774746&page=1#.TtrVVYSGKcs
Efficacy
Peter Buzing in his paper 'Comparing Different Keyboard Layouts' described five phases in the use of keyboards. These are:
Character Recognition: the phase during which the Human brain recognizes the key it must next hit
Storage Buffer: the phase during which the Human brain stores the key they have selected in their brain
Motor program: the program which searches/locates the key on the keyboard and prepares to hit it
Keystroke: the actual keystroke
Sensory Feedback: the phase in which the Human eye scans the screen and ensures that the correct key was printed.
These phases show that typing on a Keyboard can be a complex process which requires the user's full attention. Since this process is “learned” and not a natural human instinct such as speaking, it consumes the user's full attention.
Even despite the concentration that a user puts into the process, even experienced typers still produce errors.
Further, as Hick's Law predicts, an inexperienced user has a steep slope to climb before they can become efficient at Keyboard usage. Since an average keyboard offers its user 101 keys along with several secondary options (Ctrl and Fn options), it takes them a long time to realize the next key they have to hit.
Usage
Another problems that keyboard users face if switching between languages with ease. A physical keys can only carry so many printed characters. Some manufacturers provide sticker layouts for their keyboards such that one could layover their keys with a new character set, but this process is slow, and inefficient.
Other than this, keyboards also impose minimum size requirements on their manufacturers if they are to be usable. This becomes a much more serious problem on mobile devices. While some manufacturers such as RIM chooses to stick with thumb-pads, others are experimenting with on-screen and virtual keyboards.
Physical keyboards are also vulnerable to damage.
Mice
Introduction/History
Issues
Error Production
NLP
Introduction
Issues
Current Incarnations
Android Voice Recognition, Apple SIRI, VLingo
IBM Watson
Electronic Phone Operators
Competitions for NLP
Touch Interfaces
Gesture/Motion Sensors
Supplements for NLP
Augmented Reality
Survey
Methodology
Results
Appendix
Definitions
Survey Questions
Bibliography
Projects Proposals
1) Natural Language Processing
- There are several smaller ideas we could work on within this large topic such as:
- Its implementation in chat-bots?
- Sentiment analysis - data mining for financial analysis, for disease or war prevention - http://en.wikipedia.org/wiki/Sentiment_analysis
- Watson - More complicated topic, not sure what thesis we could have here? Word sense disambiguation? http://en.wikipedia.org/wiki/Word_sense_disambiguation More research needed.
- Part of speech tagging - Used on several websites in searching, tagging, merging threads or discussions : http://en.wikipedia.org/wiki/Part-of-speech_tagging
2) Kinect/Move interaction?
- Human computer interaction in video games?
- And other areas? http://www.primesense.com/en/the-living-room-experience
- Head-tracking? Gestures? http://www.youtube.com/watch?v=Jt5nhQiL4pY More research needed.
3) Usability of Social Media Tools
- How do organizations structure their tools to promote and increase user engagement
- Psychological effects of Social Media and its pull on users